boost::polygon::area returns 0 area

[dratchkov]

Hi,

I am trying to the area of small rectangular polygon. For some reason boost::polygon::area return 0. Code sample is below

#include <boost/polygon/polygon.hpp>
#include <boost/polygon/rectangle_data.hpp>
#include <boost/polygon/polygon_set_data.hpp>
#include <boost/polygon/polygon_data.hpp>
#include <boost/version.hpp>
#include <boost/format.hpp>

template<typename T>
auto make_rect(T x1, T y1, T x2, T y2) {
    auto r = boost::polygon::rectangle_data(x1, y1, x2, y2);
    boost::polygon::polygon_data<T> p;
    boost::polygon::assign(p, r);
    return p;
}

int main(int argc, char **argv) {
    std::cout << BOOST_LIB_VERSION << std::endl;
    boost::polygon::polygon_set_data<double> poly, poly1;
    poly.insert(make_rect(0.003065, 0.0007, 0.0034, 0.0009525));
    std::cout << boost::format("%e") % boost::polygon::area(poly) << std::endl;
}

The result of the above code is:

1_78
0.000000e+00

Is this a bug or I am missing a #define?

Not knowing how useful this is: I did a little bit of debugging, stepping side by side through "good" and "bad" test cases. First, "clean"-inig the polygon produces different results. Seems like cleaning is done by running a boolean op with nothing, and then this code inside boost/polygon/detail/scan_arbitrary.hpp:

    //iterator over range of vertex property elements and call result functor
    //passing edge to be output, the merged data on both sides and the result
    template <typename result_type, typename result_functor, typename iT>
    void scan(result_type& result, result_functor rf, iT begin, iT end) {
      while(begin != end) {
        x_ = (*begin).first.first.get(HORIZONTAL); //update scanline stop location
        //print_scanline();
        --x_;   // <<---- NOTE THIS
        remove_retired_edges_from_scanline();
        ++x_; // <<-- NOTE THIS
        begin = handle_input_events(result, rf, begin, end);
        remove_retired_edges_from_scanline();
        //print_scanline();
        insert_new_edges_into_scanline();
      }
      //print_scanline();
      x_ = (std::numeric_limits<Unit>::max)();
      remove_retired_edges_from_scanline();
    }

Is the code assuming that the sequence --x_ followed by ++x_ will produce x_? In the case x_ = 0.003065 the final result is 0.0030649999999999844

[dratchkov]

Should've seen this sooner in the documentation. Duh.

The coordinate data type is a template parameter of all data types and algorithms provided by the library, and is expected to be integral. Floating point coordinate data types are not supported by the algorithms implemented in the library due to the fact that the achieving floating point robustness implies a different set of algorithms and generally platform specific assumptions about floating point representations.

[dratchkov]

See previous comment